agents
llm
open-source
ai-automation
View original source
video

Agents Make All Custom Software Viable at AIE Europe

AI agents like OpenClaw turn uneconomic custom automations into reality, expanding software markets, boosting engineer demand, and enabling personal-to-enterprise scaling.

Agents Fill the Software Universe

Malte Ubl (Vercel CTO) frames AI agents as the bridge to untapped software potential. Traditional coding couldn't economically justify countless automations—business rules, if-statements, domain knowledge too bespoke or niche. Agents change this: a Venn diagram shows the full circle of 'software that should exist' now fillable, as agentic workflows handle variability without hardcoded logic.

This shifts economics: cheaper software production tests market elasticity, with no S-curve in sight. Companies pivot from SaaS buys to 'make' decisions—'SaaS apocalypse' in Silicon Valley—but demand for engineers rises as more software gets built faster. Ubl predicts juniors molded in AI will excel, while veterans adapt. Actionable archetype: start with low-hanging non-coding agents (e.g., chat SDKs hooking agents to Slack/Telegram via TypeScript bash interpreters for nanosec sandboxes) before complex coding ones.

"Agents are a new kind of software because there was always all this stuff we wanted to automate but not all of it was economically viable to do with traditional software but it is with agents." Ubl emphasizes APIs must evolve to 'AI-first' for agent consumption.

OpenClaw Ecosystem Powers Personal and Enterprise Agents

Peter Steinberger (OpenAI/OpenClaw) reports explosive growth: 'State of the Claw' highlights security bounty overload from AI-generated exploits. swyx interview reveals OpenAI's open-source stance, 'token maxing' (inflating inference to fake productivity), and 'taste' as the un-AI-able engineering moat—curated judgment on outputs.

Practical deployments proliferate:

  • Vincent Koc (Comet ML): 'Dark factories' run 60+ parallel agents overnight for codebase refactoring.
  • Radek Sienkiewicz (VelvetShark): Delegates personal life to OpenClaw via Obsidian notes, email, background tasks—keys to life handed over.
  • Onur Solmaz: ACP (Agent-Client Protocol) standardizes interactions; disposable enterprise agents on Kubernetes for isolation.
  • Fryderyk Wiatrowski (Viktor): Slack-native 'AI employee' with context across thousands of tools.
  • Merve Noyan (Hugging Face): Local coding agents via HF Hub skills for model training.

Ryan Lopopolo (OpenAI) pushes 'code is free' mindset: systems thinking + delegation spawns parallel AI coding agents. Humans steer, agents execute—harness engineering.

"Token maxing" critiqued by Gergely Orosz/swyx: Big Tech engineers waste inference padding metrics, eroding real gains.

Beyond Language: Multimodal Advances and Secure Scaling

Raia Hadsell (Google DeepMind VP Research) expands DeepMind past LLMs: Gemini Embeddings 2 for superior representations; AI cyclone prediction models; Project Genie 3 for generative skills. Non-language apps prove AI's breadth.

Security and infra harden agents:

  • Sally Ann O'Malley (Red Hat): Podman/Docker/K8s for isolation, state recovery, formal verification in Clawdbot memory architecture.
  • Nick Taylor (Pomerium): Identity-aware proxy hardens OpenClaw; live-codes MCP server from Discord.

Sunil Pai (Cloudflare): 'Code Mode' executes JavaScript in V8 isolates, bypassing slow JSON tool calls—10x speed for agent actions.

Kitze (Sizzy) roasts productivity apps, demos OS generating UI on-demand via AI. Matt Pocock (AI Hero) insists DDD/TDD combat AI 'slop': structured domains/tests filter hallucinations.

Community context: AIE grew 900% in 3 years; UK gov invests in AI infra, per Lia McBride. Event signals Europe's AI surge.

"We are speedrunning what's really an experiment in economics of how elastic the software market is." Ubl's thesis underpins agent hype.

"State of the Claw": OpenClaw's growth draws AI-generated security bounties, forcing rapid hardening.

Key Takeaways

  • Build agent archetypes today: Hook to chats (Slack/Telegram) with sandboxes; target low-hanging automations saving millions without process overhauls.
  • Delegate via systems thinking: Humans steer high-level, parallel agents execute code—'code is free.'
  • Secure OpenClaw deployments: Use Podman/K8s isolation, Pomerium proxies, ACP for standardized client-agent comms.
  • Scale with 'dark factories': Run 60+ parallel agents overnight for refactors; make disposable on K8s.
  • Personalize ruthlessly: Integrate Obsidian/email/tasks into OpenClaw for life automation.
  • Avoid token maxing: Focus on taste—curated outputs—as moat; enforce DDD/TDD against slop.
  • Upgrade tools: Adopt Code Mode (V8 JS execution), HF local agents, Gemini Embeddings for non-LLM tasks.
  • Design AI-first APIs: Agents as primary users demand structured, predictable interfaces.
  • Monitor economics: Track if cheaper software expands markets—demand for AI engineers follows.
Video description
Timestamps 00:13:10 - Opening remarks by Phil Hawksworth 00:21:26 - Lia McBride (AI Engineer): AIE community's 900% growth and UK gov's AI infrastructure investment. 00:24:25 - Malte Ubl (Vercel): AI agents as the new application layer, custom automation now economically viable, APIs must be AI first. 00:42:39 - Raia Hadsell (Google DeepMind): DeepMind beyond language: Gemini Embeddings 2, AI cyclone prediction, and Project Genie 3. 01:07:08 - Ryan Lopopolo (OpenAI): "Code is free", systems thinking and delegation for parallel AI coding agents. 01:25:48 - Peter Steinberger (OpenAI): "State of the Claw": OpenClaw's explosive growth and AI generated security bounty overload. 01:45:12 - Break: Morning coffee 02:28:13 - swyx with Peter Steinberger (OpenAI): OpenAI on open source, "token maxing," and "taste" as the ultimate engineering moat. 02:55:01 - Vincent Koc (Comet ML): "Dark factories": 60+ parallel AI agents on overnight codebase refactoring. 03:14:07 - Radek Sienkiewicz (VelvetShark): Handing personal life to OpenClaw via Obsidian, email, and background tasks. 03:34:12 - Sally Ann O'Malley (Red Hat): Secure agent deployment with Podman, Docker, and K8s, isolation and state recovery. 03:57:05 - Nick Taylor (Pomerium): Hardening OpenClaw with Pomerium's identity aware proxy and live coding an MCP server from Discord. 04:14:35 - Break: Lunch 05:41:51 - Onur Solmaz (OpenClaw): ACP for standardized agent-to-client interactions and disposable enterprise agents on K8s. 06:02:17 - Merve Noyan (Hugging Face): HF ecosystem for local coding agents and model training via Hub skills. 06:22:36 - Fryderyk Wiatrowski (Viktor): "Viktor", Slack native AI employee with context across thousands of integrated tools. 06:42:09 - Break: Afternoon 07:42:39 - Gergely Orosz (The Pragmatic Engineer) with swyx: "Token Maxing", big tech engineers wasting AI inference to inflate productivity metrics. 08:09:26 - Kitze (Sizzy): Modern productivity apps roasted, and an OS where AI generates UI on demand. 08:29:42 - Matt Pocock (AI Hero): Why DDD and TDD matter more than ever against AI generated "slop." 08:48:31 - Sunil Pai (Cloudflare): "Code Mode", LLMs executing JavaScript in V8 isolates, bypassing slow JSON tool calls. 09:07:04 - Closing remarks by Phil Hawksworth April 9, 2026 - all times in GMT+1 (UK Time) EVENT OPENER 8:54am - Stream beings 9:00am - Event Opener - opening video, MC, AIE welcome from Lia McBride, General Manager AIE Welcome | AIE team, AI Engineer OPENING KEYNOTES * 9:10am - Keynote Malte Ubl | CTO, Vercel * 9:30am - Keynote Raia Hadsell | VP of Research, Google DeepMind * 9:50am - Harness Engineering: How to Build Software When Humans Steer and Agents Execute Ryan Lopopolo | Member of Technical Staff, OpenAI * 10:10am - OpenClaw update Peter Steinberger | OpenClaw, OpenAI OPENCLAW/PERSONAL AGENTS BREAKOUT 1 * 11:15am - OpenClaw AMA Peter Steinberger | OpenClaw, OpenAI Gergely Orosz | Moderator, The Pragmatic Engineer * 11:40am - Breakout talk Vincent Koc | AI Research Engineer, Evals, DevRel, Comet ML - OpenClaw * 12:00pm - I Gave an AI Agent the Keys to My Life (Here's What Happened) Radek Sienkiewicz | Developer Relations, LayerZero * 12:20pm - Proving Agents Correct: Memory Architecture and Formal Verification for Clawdbot Sally OMalley | Principal Software Engineer, Red Hat * 12:40pm - Breakout talk Nick Taylor | Developer Advocate, Pomerium OPENCLAW/PERSONAL AGENTS BREAKOUT 2 Igor Karpovich | Senior Principal Engineer, Skyscanner * 2:30pm - Scaling Agents on Kubernetes with acpx and ACP Onur Solmaz | OpenClaw Maintainer, OpenClaw * 2:50pm - What's new in AI Audio? Thor Schaeff | Developer Relations Engineer, Google DeepMind * 3:10pm - Viktor — AI Coworker That Lives in Slack Fryderyk Wiatrowski | CEO, Viktor CLOSING KEYNOTES * 4:30pm - Software Engineering + AI = ? Gergely Orosz | Author, The Pragmatic Engineer swyx | Founder, AI Engineer * 5:00pm - Keynote Kitze | AIE Top Speaker, Sizzy.co * 5:20pm - It Ain't Broke: Why Software Fundamentals Matter More Than Ever Matt Pocock | AI Hero, Total TypeScript * 5:40pm - Keynote Sunil Pai | Principal Systems Engineer, Cloudflare```

Summarized by x-ai/grok-4.1-fast via openrouter

8151 input / 2491 output tokens in 21715ms

© 2026 Edge